This repository serves as a curated collection of reading notes and resources on domain generalization, domain adaptation, causality, robustness, prompt learning, optimization, and generative models. It is primarily aimed at researchers and practitioners in machine learning seeking to stay updated with the latest advancements and understand the theoretical underpinnings of these complex topics. The collection offers a structured overview of key papers, categorized by year and sub-topic, with direct links to code and detailed reading notes.

How It Works

The repository organizes research papers by key areas such as Generalization/OOD, LLM safety, Test-time adaptation, Robustness/Adaptation/Fairness/OOD Detection, Data-Centric/Prompt/Large-Pretrain-Model, Optimization/GNN/Energy/Causality/Others. Each entry typically includes a brief description of the paper's contribution, its venue (e.g., NeurIPS, ICML, CVPR), and links to associated code and the author's personal reading notes. This structure facilitates a deep dive into specific research directions and allows users to track the evolution of ideas within these fields.

Quick Start & Requirements

This repository is a collection of research papers and reading notes, not a runnable software package. No installation or execution commands are applicable. The primary requirement is access to academic papers and a willingness to engage with technical content.

Highlighted Details

• Extensive coverage of papers from major ML conferences (NeurIPS, ICML, CVPR) from 2017 to late 2023.
• Detailed reading notes provided for many papers, offering insights beyond the abstract.
• Categorization includes emerging areas like LLM safety and specific techniques like Test-Time Adaptation.
• Links to code repositories for many papers enable practical exploration and reproduction.

Maintenance & Community

The repository is maintained by yfzhang114, a PhD student with research experience at Microsoft and Alibaba DAMO Academy. The project is actively updated, with recent additions reflecting papers from NeurIPS 2023 and LLM safety research. The author encourages collaboration and discussion via their personal homepage.

Licensing & Compatibility

The repository itself does not host code that requires a specific license; it is a collection of links and notes. The licensing of the individual papers and their associated code repositories would need to be checked on a per-item basis. Compatibility for commercial use or closed-source linking depends entirely on the licenses of the linked papers and their code.

Limitations & Caveats

This is a personal collection of reading notes and does not represent a comprehensive or exhaustive survey of all research in these fields. The depth of coverage and the selection of papers reflect the author's personal research interests and may not cover all relevant work. The "reading notes" are subjective and represent the author's interpretation.